Promptos Starter

Deploy your own AI-powered content platform in 5 minutes. No AI experience needed.

Promptos Starter is a ready-to-use template for building AI content generation apps. It comes with a Task Breakdown Engine, 4 built-in modules, and a visual playground — all configurable through simple JSON and Markdown files.

Zero coding required. Edit a config file, write your prompts in Markdown, click deploy.

English | 中文版本

Demo

You: "I want to build a SaaS product for small businesses"

Promptos (Task Breakdown Engine):
├── Phase 1: Research & Define (Day 1-3)
│   ├── Identify target customer segments
│   ├── Analyze 3 competing products
│   └── Define core feature set (max 3 features)
├── Phase 2: Design & Prototype (Day 4-6)
│   ├── Create user flow diagrams
│   ├── Build clickable prototype
│   └── Validate with 5 target users
├── Phase 3: MVP Development (Day 7-12)
│   ├── Set up tech stack (Next.js + Vercel)
│   ├── Build authentication + core features
│   └── Deploy staging environment
└── Phase 4: Launch & Iterate (Day 13-15)
    ├── Beta testing with 10 users
    ├── Fix critical issues
    └── Public launch + marketing

Features

1 Core Engine + 4 Modules = Complete Workflow

┌─────────────────────────────────────────────────────────┐
│                                                         │
│   📥 Input: "Build an online education platform"        │
│                                                         │
│   ┌─────────────────────────────────────────────┐       │
│   │  Task Breakdown Engine                      │       │
│   │  Decomposes any goal into actionable steps  │       │
│   └──────────────────┬──────────────────────────┘       │
│                      │                                  │
│          ┌───────────┼───────────┐                      │
│          ▼           ▼           ▼                      │
│   ┌───────────┐ ┌─────────┐ ┌──────────┐               │
│   │ PM / OKR  │ │  Deep   │ │  Tech    │               │
│   │ Planner   │ │Analysis │ │  Stack   │               │
│   └─────┬─────┘ └────┬────┘ └────┬─────┘               │
│         │            │           │                      │
│         ▼            ▼           ▼                      │
│   OKR + Timeline  Risk Report  Architecture             │
│                                                         │
│         └────────────┼───────────┘                      │
│                      ▼                                  │
│              ┌──────────────┐                           │
│              │Writing Master│                           │
│              │  Final Docs  │                           │
│              └──────────────┘                           │
│                                                         │
└─────────────────────────────────────────────────────────┘

Component	What It Does	Example Output
Task Breakdown Engine	Decomposes any complex goal into phases, tasks, and sub-tasks with dependencies	Project plan with 4 phases, 12 tasks
PM / OKR Planner	Creates OKRs, milestones, and execution timeline from a task list	Q1 OKR with 3 objectives, 9 key results
Deep Analysis	MECE analysis of any topic — risks, opportunities, tradeoffs	5-dimension analysis with recommendations
Tech Stack Advisor	Recommends architecture and tech choices for your project	Stack comparison table + architecture diagram
Writing Master	Generates polished documents, reports, and content	Professional report with structure and formatting

Multi-LLM Support

Switch between AI providers by changing one line:

# Use DeepSeek (default, cheapest)
DEEPSEEK_API_KEY=sk-xxx

# Or use Google Gemini
GEMINI_API_KEY=xxx

# Or use both — switch per request

Supports any OpenAI-compatible API (DeepSeek, Groq, Together, Ollama, etc.) + Google Gemini. Add a new provider in 30 seconds — just edit config/providers.json.

Visual Playground

Every deployment includes a built-in playground UI:

Select engine or module from dropdown
Type your input
See AI output in real-time
Copy the API call as cURL

No Postman, no terminal. Test everything in your browser.

Quick Start

Option A: One-Click Deploy (Recommended)

Click the button above
Enter your API key (get one from DeepSeek — it's $0.14/million tokens)
Wait 60 seconds for deployment
Open your app and start using it

Option B: Run Locally

# Clone
git clone https://github.com/eiragu/promptos-starter.git
cd promptos-starter

# Install
npm install

# Configure
cp .env.example .env.local
# Edit .env.local — add your API key

# Run
npm run dev

Open http://localhost:3000 — done.

Customization

Add a New Prompt Module (2 minutes, no coding)

Step 1: Create a Markdown file for your prompt template:

<!-- config/prompts/modules/my-module.md -->
# Social Media Post Generator

You are an expert social media content creator.

## Task
Generate engaging social media posts based on the user's input.

## Requirements
- Platform-optimized (Twitter: 280 chars, LinkedIn: professional tone)
- Include 3 variations
- Suggest relevant hashtags

## Output Format
### Post 1 (Twitter)
...
### Post 2 (LinkedIn)
...
### Post 3 (Instagram)
...

Step 2: Register it in config/modules.json:

{
  "id": "social_media",
  "label": "Social Media Generator",
  "labelCN": "社媒内容生成",
  "category": "marketing",
  "prompt": "modules/my-module"
}

Step 3: Done. Your new module is live — available in the playground and via API.

Add a New LLM Provider (30 seconds)

Edit config/providers.json:

{
  "groq": {
    "label": "Groq (Llama 3)",
    "type": "openai-compatible",
    "baseURL": "https://api.groq.com/openai/v1",
    "model": "llama-3.3-70b-versatile",
    "envKey": "GROQ_API_KEY"
  }
}

Add GROQ_API_KEY=xxx to your .env.local. That's it.

Customize Existing Prompts

All prompts live in config/prompts/ as Markdown files. Open any file, edit the text, save. Changes take effect immediately in development mode.

config/prompts/
├── core/
│   └── task-breakdown.basic.md    ← Edit the core engine prompt
└── modules/
    ├── writing-master.md          ← Edit any module prompt
    ├── deep-analysis.md
    ├── pm-okr.md
    └── tech-stack.md

API Reference

POST `/api/core/run` — Run Core Engine

curl -X POST https://your-app.vercel.app/api/core/run \
  -H "Content-Type: application/json" \
  -d '{
    "coreKey": "task_breakdown",
    "tier": "basic",
    "userInput": "Build a mobile fitness app",
    "engineType": "deepseek"
  }'

Response:

{
  "ok": true,
  "output": "## Project Breakdown\n\n### Phase 1: Research...",
  "text": "## Project Breakdown\n\n### Phase 1: Research...",
  "content": "## Project Breakdown\n\n### Phase 1: Research...",
  "meta": {
    "requestId": "a1b2c3d4",
    "coreKey": "task_breakdown",
    "tier": "basic",
    "engineType": "deepseek"
  }
}

POST `/api/run` — Run Any Module

curl -X POST https://your-app.vercel.app/api/run \
  -H "Content-Type: application/json" \
  -d '{
    "moduleId": "writing_master",
    "userInput": "Write a project proposal for an AI chatbot",
    "engineType": "deepseek"
  }'

GET `/api/registry` — List Available Modules

curl https://your-app.vercel.app/api/registry

GET `/api/ping` — Health Check

curl https://your-app.vercel.app/api/ping
# → { "ok": true, "timestamp": "2026-03-29T..." }

Project Structure

promptos-starter/
├── promptos.config.json        ← Main config (start here)
├── config/
│   ├── providers.json          ← LLM providers (add/remove models)
│   ├── engines.json            ← Core engine definitions
│   ├── modules.json            ← Module registry (add your own!)
│   └── prompts/                ← All prompts as Markdown files
│       ├── core/
│       │   └── task-breakdown.basic.md
│       └── modules/
│           ├── writing-master.md
│           ├── deep-analysis.md
│           ├── pm-okr.md
│           └── tech-stack.md
├── src/
│   ├── app/
│   │   ├── page.tsx            ← Playground UI
│   │   └── api/                ← API routes (auto-configured)
│   └── lib/
│       ├── provider.ts         ← Multi-LLM unified interface
│       ├── engines/            ← Core engine logic
│       ├── modules/            ← Module resolution
│       └── middleware/         ← Error handling, auth, tracing
├── .env.example
├── package.json
└── README.md

You only need to touch the config/ folder. Everything in src/ works automatically.

Comparison

Feature	Promptos Starter	LangChain	Raw OpenAI SDK
Time to deploy	5 min	Hours	Hours
Coding required	No	Yes (Python)	Yes
Multi-LLM support	Built-in	Plugin-based	Manual
Visual playground	Included	No	No
Prompt management	Markdown files	Code	Code
One-click deploy	Vercel button	No	No
Production-ready	Yes	Depends	No

Full Architecture

Want to understand how Promptos works under the hood?

Check out the complete architecture documentation: Promptos Architecture — covers the design decisions, patterns, and lessons learned from building a production AI platform with 80+ modules.

Roadmap

Task Breakdown Engine (core)
4 built-in modules (Writing, Analysis, PM/OKR, Tech Stack)
Multi-LLM support (DeepSeek, Claude, Gemini, Kimi, OpenRouter, SiliconFlow)
Visual Playground UI
One-click Vercel deploy
Streaming output support
Module marketplace (community prompts)
Pro engine tier (deeper analysis, longer output)
Plugin system (pre/post processing hooks)

Who Built This

Eira — 10 years in financial sales, now building AI products.

Promptos Starter is extracted from FuYouAI, a production AI content generation platform with 80+ modules serving real users. This isn't a toy project — it's battle-tested code from a live product.

FuYouAI — fuyouai.com (the full platform)
Architecture Docs — promptos-architecture
GitHub — @eiragu

Support This Project

If Promptos Starter helps you build something, consider:

Giving this repo a Star — it helps others discover the project
Sharing your creation — tag me on Twitter/X or open a Discussion
Contributing — prompt templates, bug fixes, translations are all welcome
Checking out FuYouAI — the full platform with 80+ AI modules

License

MIT — use it for anything, commercial or personal.

中文版本